An image searching method includes: resizing an input image in question; generating 3D image identifiers for the resized input image in question; and performing an image search for the input image in question by using the 3D image identifiers. Said resizing an input image includes: extracting a black-and-white image and a color image from the input image in question; resizing the black-and-white image; and resizing the color image. Said generating 3D image identifiers includes: extracting an MGST feature of the input image in question; extracting an angular partition feature of the input image in question; and extracting a color feature of the input image in question. Further, said performing an image search includes: calculating similarity between representative colors of the input image in question and a reference image; and if the similarity is above a predetermined level, matching the 3D image identifiers for the two images.
Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. An image searching method, comprising: resizing an input image in question; generating 3D image identifiers for each pixel from the resized input image in question, the 3D image identifiers being representative of a common feature of transformation of the input image; performing an image search by comparing a reference image with the input image in question using the 3D image identifiers; wherein said generating the 3D image identifiers includes: extracting a Modified Generalized Symmetry Transform (MGST) feature of the input image in question; extracting an angular partition feature of the input image in question; and extracting a color feature of the input image in question.
An image searching method resizes an input image, generates 3D image identifiers for each pixel of the resized image representing a common feature of transformation, and performs an image search by comparing a reference image with the input image using these identifiers. The 3D image identifiers are created by extracting a Modified Generalized Symmetry Transform (MGST) feature, an angular partition feature, and a color feature from the input image.
2. The image searching method of claim 1 , wherein said resizing the input image includes: extracting a black-and-white image and a color image from the input image in question; resizing the black-and-white image; and resizing the color image.
The image searching method from the previous description resizes the input image by first extracting a black-and-white and a color image. It then resizes both the black-and-white and the color image separately before proceeding to generating 3D image identifiers for each pixel of the resized image representing a common feature of transformation, and performing an image search by comparing a reference image with the input image using these identifiers. The 3D image identifiers are created by extracting a Modified Generalized Symmetry Transform (MGST) feature, an angular partition feature, and a color feature from the input image.
3. The image searching method of claim 2 , wherein the black-and-white image is resized by bi-cubic interpolation.
The image searching method from the previous description resizes the input image by first extracting a black-and-white and a color image. It then resizes both the black-and-white and the color image separately before proceeding to generating 3D image identifiers for each pixel of the resized image representing a common feature of transformation, and performing an image search by comparing a reference image with the input image using these identifiers. The 3D image identifiers are created by extracting a Modified Generalized Symmetry Transform (MGST) feature, an angular partition feature, and a color feature from the input image. The black and white image is resized using bi-cubic interpolation.
4. The image searching method of claim 2 , wherein said resizing the color image includes: dividing the input image in question into three channel images; resizing each of the three channel images by bi-cubic interpolation; and matching the resized three channel images with each other again to resize the color image.
The image searching method from claim 2 resizes the color image by dividing it into three channel images, resizing each channel using bi-cubic interpolation, and then recombining the resized channels. The method first extracts a black-and-white and a color image. It then resizes both the black-and-white and the color image separately before proceeding to generating 3D image identifiers for each pixel of the resized image representing a common feature of transformation, and performing an image search by comparing a reference image with the input image using these identifiers. The 3D image identifiers are created by extracting a Modified Generalized Symmetry Transform (MGST) feature, an angular partition feature, and a color feature from the input image.
5. The image searching method of claim 1 , wherein said extracting the MGST feature includes: classifying pixel pairs which are symmetrically disposed about the central pixel in a certain area within the input image in question; and accumulating a degree of symmetry for each of the pixel pairs to calculate the MGST feature.
The image searching method resizes an input image, generates 3D image identifiers for each pixel of the resized image representing a common feature of transformation, and performs an image search by comparing a reference image with the input image using these identifiers. To extract the Modified Generalized Symmetry Transform (MGST) feature, the method classifies pixel pairs symmetrically disposed around a central pixel within a specific area of the input image and calculates the MGST feature by accumulating a symmetry degree for each pair.
6. The image searching method of claim 5 , wherein the MGST feature is divided by quantization in 7 levels.
The image searching method resizes an input image, generates 3D image identifiers for each pixel of the resized image representing a common feature of transformation, and performs an image search by comparing a reference image with the input image using these identifiers. To extract the Modified Generalized Symmetry Transform (MGST) feature, the method classifies pixel pairs symmetrically disposed around a central pixel within a specific area of the input image and calculates the MGST feature by accumulating a symmetry degree for each pair. The MGST feature is then quantized into 7 levels.
7. The image searching method of claim 1 , wherein the angular partition feature is divided by quantization in 13 levels.
The image searching method resizes an input image, generates 3D image identifiers for each pixel of the resized image representing a common feature of transformation, and performs an image search by comparing a reference image with the input image using these identifiers. The angular partition feature is divided by quantization in 13 levels.
8. The image searching method of claim 1 , wherein said extracting the color feature includes: obtaining an average RGB value of a 3×3 area about each pixel of the input image in question; and segmenting the average RGB value for mapping to a hue value in hue saturation intensity color space.
The image searching method resizes an input image, generates 3D image identifiers for each pixel of the resized image representing a common feature of transformation, and performs an image search by comparing a reference image with the input image using these identifiers. Extracting the color feature involves obtaining an average RGB value from a 3x3 area around each pixel and segmenting this average to map it to a hue value within the hue saturation intensity color space.
9. The image searching method of claim 8 , wherein the hue value is divided by quantization in 10 levels.
The image searching method resizes an input image, generates 3D image identifiers for each pixel of the resized image representing a common feature of transformation, and performs an image search by comparing a reference image with the input image using these identifiers. Extracting the color feature involves obtaining an average RGB value from a 3x3 area around each pixel and segmenting this average to map it to a hue value within the hue saturation intensity color space. The hue value is quantized into 10 levels.
10. The image searching method of claim 1 , wherein said performing the image search includes: calculating a similarity between representative colors of the input image in question and the reference image; and if the similarity is above a predetermined level, matching the 3D image identifiers for each of the representative colors of the input image in question and the reference image.
The image searching method resizes an input image, generates 3D image identifiers for each pixel of the resized image representing a common feature of transformation, and performs an image search by comparing a reference image with the input image using these identifiers. The image search involves calculating a similarity between representative colors of the input image and a reference image. If the similarity exceeds a threshold, the method proceeds to match the 3D image identifiers for each of the representative colors of the input image and the reference image.
11. The image searching method of claim 10 , wherein the representative colors of the input image in question are composed of 5 representative colors selected out of colors extracted from the input image in question as the color feature.
The image searching method resizes an input image, generates 3D image identifiers for each pixel of the resized image representing a common feature of transformation, and performs an image search by comparing a reference image with the input image using these identifiers. The image search involves calculating a similarity between representative colors of the input image and a reference image. If the similarity exceeds a threshold, the method proceeds to match the 3D image identifiers for each of the representative colors of the input image and the reference image. The representative colors consist of 5 colors selected from colors extracted from the input image as the color feature.
12. The image searching method of claim 10 , wherein said matching the 3D image identifiers includes: extracting the 3D image identifiers of each of the input image in question and the reference image; and calculating a distance between 3D image identifiers of pixels for the 3D image identifiers of each of the two images for matching there between.
The image searching method resizes an input image, generates 3D image identifiers for each pixel of the resized image representing a common feature of transformation, and performs an image search by comparing a reference image with the input image using these identifiers. The image search involves calculating a similarity between representative colors of the input image and a reference image. If the similarity exceeds a threshold, the method proceeds to match the 3D image identifiers for each of the representative colors of the input image and the reference image. Matching the 3D image identifiers includes extracting the 3D identifiers from both images and calculating a distance between pixel 3D identifiers to match the images.
13. The image searching method of claim 12 , wherein the 3D image identifiers are configured in a 3D histogram structure.
The image searching method resizes an input image, generates 3D image identifiers for each pixel of the resized image representing a common feature of transformation, and performs an image search by comparing a reference image with the input image using these identifiers. The image search involves calculating a similarity between representative colors of the input image and a reference image. If the similarity exceeds a threshold, the method proceeds to match the 3D image identifiers for each of the representative colors of the input image and the reference image. The 3D image identifiers are organized in a 3D histogram structure.
14. The image searching method of claim 1 , wherein the input image in question is a user created contents image.
The image searching method resizes an input image, generates 3D image identifiers for each pixel of the resized image representing a common feature of transformation, and performs an image search by comparing a reference image with the input image using these identifiers. The input image is a user created content image.
15. A non-transitory computer-readable storage medium including computer executable instructions, wherein the instructions, when executed by a processor, cause the processor to perform an image searching method, the method comprising: resizing a size of an input image in question; generating 3D image identifiers from the resized input image in question, the 3D image identifiers being representative of a common feature of transformation of the input image; performing an image search by comparing a reference image with the input image in question using the 3D image identifiers; wherein said generating the 3D image identifiers includes: extracting a Modified Generalized Symmetry Transform (MGST) feature of the input image in question; extracting an angular partition feature of the input image in question; and extracting a color feature of the input image in question.
A non-transitory computer-readable storage medium contains instructions for an image searching method. When executed, the instructions resize an input image, generate 3D image identifiers for each pixel representing a common feature of transformation, and perform an image search by comparing a reference image with the input image using these identifiers. The 3D image identifiers are created by extracting a Modified Generalized Symmetry Transform (MGST) feature, an angular partition feature, and a color feature from the input image.
16. The non-transitory computer-readable storage medium of claim 15 , wherein said resizing to the input image includes: extracting a black-and-white image and a color image from the input image in question; resizing the black-and-white image; and resizing the color image.
A non-transitory computer-readable storage medium contains instructions for an image searching method. When executed, the instructions resize an input image by extracting a black-and-white and a color image and resizing them separately, generate 3D image identifiers for each pixel representing a common feature of transformation, and perform an image search by comparing a reference image with the input image using these identifiers. The 3D image identifiers are created by extracting a Modified Generalized Symmetry Transform (MGST) feature, an angular partition feature, and a color feature from the input image.
17. The non-transitory computer-readable storage medium of claim 15 , wherein said performing the image search includes: calculating similarity between representative colors of the input image in question and a reference image; and if the similarity is above a predetermined level, match the 3D image identifiers for the two images.
A non-transitory computer-readable storage medium contains instructions for an image searching method. When executed, the instructions resize an input image, generate 3D image identifiers for each pixel representing a common feature of transformation, and perform an image search by comparing a reference image with the input image using these identifiers. The image search involves calculating the similarity between representative colors of the input image and a reference image. If the similarity exceeds a threshold, match the 3D image identifiers for the two images.
18. The non-transitory computer-readable storage medium of claim 15 , wherein the 3D image identifiers are configured in a 3D histogram structure.
A non-transitory computer-readable storage medium contains instructions for an image searching method. When executed, the instructions resize an input image, generate 3D image identifiers for each pixel representing a common feature of transformation, and perform an image search by comparing a reference image with the input image using these identifiers. The 3D image identifiers are configured in a 3D histogram structure.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
December 29, 2009
August 27, 2013
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.